artificial neural network
artificial neural network
(artificial intelligence)A neural network is a processing device, either analgorithm, or actual hardware, whose design was inspired bythe design and functioning of animal brains and componentsthereof.
Most neural networks have some sort of "training" rule wherebythe weights of connections are adjusted on the basis ofpresented patterns. In other words, neural networks "learn"from examples, just like children learn to recognise dogs fromexamples of dogs, and exhibit some structural capability forgeneralisation.
Neurons are often elementary non-linear signal processors (inthe limit they are simple threshold discriminators). Anotherfeature of NNs which distinguishes them from other computingdevices is a high degree of interconnection which allows ahigh degree of parallelism. Further, there is no idle memorycontaining data and programs, but rather each neuron ispre-programmed and continuously active.
The term "neural net" should logically, but in common usagenever does, also include biological neural networks, whoseelementary structures are far more complicated than themathematical models used for ANNs.
See Aspirin, Hopfield network, McCulloch-Pitts neuron.
Usenet newsgroup: news:comp.ai.neural-nets.